Summary
Shubham Phal is a Senior Research Engineer at DeepMind in Mountain View with eight years of experience building and shipping 0→1 AI products and research systems. He’s served as tech lead across multimodal alignment, model behavior, and education-focused initiatives for the Gemini family, driving projects like LearnLM, Ask YouTube, HallMate, and the cross-org Project Astra Tutor that were demoed at Google I/O. Comfortable as an individual contributor who still codes, he combines hands-on RL and multimodal research with program-level coordination—once rallying 100+ contributors to deliver a tutor product in ~4 months. His background spans applied research internships (Microsoft Research, Amazon) and an MS from Carnegie Mellon, where he contributed to top-ranked dialog systems for the Alexa Prize. Now in DeepMind’s AGI Alignment team, he’s developing new RL algorithms for reasoning over images and audio while translating research into real-world AI tutors and learning tools.
8 years of coding experience
3 years of employment as a software developer
Bachelor's degree, Bachelor's degree at R. V. College of Engineering, Bangalore
MS Computational Data Science, School of Computer Science, MS Computational Data Science, School of Computer Science at Carnegie Mellon University
Konkani, Marathi, Hindi, English